Changes in tissue fatty acid composition in murine malignancy and following anticancer therapy.

We studied the mouse NC tumour, a subcutaneously transplanted adenocarcinoma originally of mammary origin. Measurements per g tissue were made of 17 fatty acids (FAs), the combined amounts of n-3, n-6, saturated, unsaturated, and total FAs, and of various FA ratios in the tumour, mammary tissue, spleen, liver and plasma. Compared with mammary tissue from normal mice, tumours of vehicle-treated controls had less of seven of the FAs and more of two FAs. Mice bearing the NC tumour often had changed (usually decreased) amounts of FAs in the 'normal' spleen, liver and plasma, but not in mammary tissue. Treatment with methotrexate (MTX) was studied alone and with indomethacin which can potentiate MTX cytotoxicity. Indomethacin 1.25 mg kg-1 (INDO) increased the amounts of 3/17 tumours FAs and the unsaturated FAs, but reduced 9/17 FAs, the saturated and the unsaturated FAs in 'normal' mammary tissue, and usually had no effect on the FAs of other tissues. MTX 2 or 4 mg kg-1 (MTX 2 or 4 mg) +/- INDO in general partly restored (increased) the amounts of tumour FAs, and reduced the saturated/unsaturated FA ratio. In the 'normal' spleen and plasma also, but not in the liver, MTX 2 mg generally somewhat restored the FA composition. However, as in the liver, the spleen 20:4 and 22:6 (which form prostaglandins and lipid peroxides) did not increase in the presence of INDO. With MTX 4 mg, some of the plasma and liver FAs decreased, in contrast to the tumour. There was generally no evidence of MTX potentiation by INDO. These results are discussed in relation to carcinogenesis, cachexia, and the response to treatment.

Relationships between lipids and cancer are not fully understood. Some epidemiological studies suggest the involvement of dietary fats in human cancer development (Correa, 1981;Holm et al., 1989;Prentice et al., 1989;Young & Young, 1989); both the quality and quantity of dietary fats might influence tumour incidence. In animal studies, linoleic acid (an n-6 FA) promoted tumour growth and development, with concomitant increases of eicosanoid synthesis and cell division, and depression of the immune response (Karmali, 1987). Conversely, diets rich in n-3 FAs inhibited some cancers, possibly by decreasing arachidonate metabolism (Karmali, 1987;Abou-EI-Ela et al., 1988).
Cancer cachexia, the weight loss that can accompany malignancy, involves gross metabolic disturbance. In mice, this was reduced by dietary manipulation with fish oil (Tisdale & Dhesi, 1990) or by treatment with indomethacin (Gelin et al., 1991). FA changes seen in our experiments might be relevant to this condition.
Our research into methotrexate (MTX) started because we found that the cyclo-oxygenase inhibitors flurbiprofen and indomethacin (INDO) decreased cancer development and spread (Bennett et al., 1979. We then demonstrated that INDO potentiates the anticancer effect of MTX in vitro and in vivo. The mechanism is not clear, but the effect in vitro probably does not involve MTX displacement from binding sites on serum proteins, or inhibition of prostaglandin formation, cAMP phosphodiesterase or of calcium transport (Gaffen et al., 1985(Gaffen et al., , 1989Bennett et al., 1987;Gaffen et al., 1991). Possibilities examined in the present study are whether INDO and MTX alone and together affect the fatty acids (FAs) of malignant and 'normal' tissues, and whether the potentiation of MTX cytotoxicity involves alteration of tumour FA composition. We have therefore measured various FAs in extracts of mouse NC tumour, mammary tissue, spleen, liver and plasma, and the effects of MTX and INDO on them.

Materials and methods
Mouse treatment in vivo The NC carcinoma used in these studies arose initially in the mammary region of a WHT/Ht mouse (Hewitt et al., 1976) and has been transplanted in the same strain since then. Metastasis to the lungs and mediastinum, local lymphatic spread and recurrence in the excision scar commonly occur.
Female WHT/Ht mice aged 2-4 months and weighing 24-27 g at the start of the experiment were fed SDS No. 1 modified diet (Special Diet Services Ltd., Essex, UK) and had free access to water. They were weighed at intervals of 2-4 days starting 10 days before tumour transplantation; during this short experiment there were no significant differences between the groups. The two separate experiments resulted in combined numbers of six to nine mice in each of the seven groups. On day 0 all but one group of mice were injected s.c. into the left flank with approximately 106 NC carcinoma cells (Bennett et al., 1979. By day 8, 80% of the tumours were palpable; by day 11 all the mice had palpable and visible tumours. On days 15 -18, the six tumour-bearing groups received orally administered vehicle (syrup) alone or containing MTX 2 or 4 mg kg-' (MTX 2 or 4 mg), INDO 1.25 mg kg-' (INDO) alone, or MTX 2 or 4 mg with INDO. A control group without tumour received only the syrup vehicle.
On day 18, 2.5-7 h after the last drug administration, the mice were anaesthetised with ether, blood was withdrawn by cardiac puncture into a tube containing 50 units of heparin, and the plasma obtained after centrifugation (1,500 g 4°C, 10 min). The mice were killed by cervical dislocation, and the transplanted tumours, liver, spleen, and mammary tissue Correspondence: A. Bennett. *Present address: Department of Pharmacology, Cerrahpasa Faculty of Medicine, University of Istanbul, Istanbul, Turkey. excised, weighed, and frozen at -70°C for 1 week prior to FA analysis. The 'normal' tissues were all macroscopically free of tumour.
Tissue homogenisation The frozen tissue was thawed but kept cold in bottles on ice. Carefully weighed tissue (100-200 mg) was cut into small pieces, homogenised (100 mg ml'; cold 154 mM NaCl; 30 s; Silverson homogeniser) and 1 ml of homogenate was removed for lipid extraction.

Total lipid extraction
The total lipids were extracted according to the method of Folch et al. (1957). Briefly, to 1 ml tissue homogenate or plasma were added 2 ml methanol, 100 p1 internal standard (10-I00 Ag heptadecanoic acid in chloroform), and 3.9 ml chloroform. The mixture was vortex-mixed for 1 min, centrifuged (2,000 g, O min, 4°C), and the chloroform phase was removed and evaporated to dryness under a stream of nitrogen at 37°C. After dissolving the extract in di-isopropyl ether/1-butanol (6:4, 2 ml), 1 ml of 50 mM NaCl was added, vortexed-mixed and centrifuged (2,000g, 10min, 4C). The upper organic phase containing the total lipids was evaporated to dryness under nitrogen at 37°C.

Results
Tissue weights NC tumours from untreated mice weighed 794 mg (715-1,000) at day 18. Treatment with MTX 4 mg kg-' (MTX 4 mg) alone or with indomethacin 1.25 mg kg-' (INDO) reduced the tumour weights by 44 and 57% respectively, whereas INDO alone or with MTX 2 mg had little or no effect (Table   I).
At day 18 the spleens from normal mice given vehicle weighed 72 mg (30-110), whereas those from mice with untreated tumours were 85% heavier (P <0.003). Treatment with MTX 4 mg + INDO decreased the spleen weight to 109 and 85 mg (51 and 18% respectively more than in normal mice). MTX 2 mg ± INDO tended to reduce the spleen weight, but INDO alone had no effect ( Table I).
The weight of livers from normal mice was 1.21 g (1.11-1.32), about the same as in the cancer-bearing groups, and was unchanged by drug treatment.
Fatty acid changes Since there are seven groups each with measurements of 17 FAs, combined amounts of n-3, n-6, saturated, unsaturated and total FAs, and calculations of various ratios, it is to be expected that by chance some analyses will indicate a statistically significant difference when none really exists (a Type I error). Nevertheless, it seems that at least some of the treatments resulted in genuine changes. Because of the large amount of data, we have selected for discussion the aspects mainly related to tumour FAs, the effects of the tumour on normal tissue FAs, and to a possible MTX/INDO interaction. Details of all aspects are presented in the Tables.
The FAs in 'normal' tissues from tumour-bearing vehicletreated mice are compared with normal controls (i.e. no cancer) that received only vehicle. FAs in the drug-treated groups are compared with tissues from vehicle-treated tumour-bearing mice.
Tumour fatty acids Table II shows the amounts g-' of FAs in the total tumour Statistics lipids.
Results are presented as median values and interquartile ranges or as per cent median changes. The Mann-Whitney U-test (2-tailed) was used for comparisons of FA content. Only P values of at most 0.1 are shown in the tables, and unqualified statements in the text imply a P value of at most 0.05. All doses are mg kg-'; for simplicity this is usually shortened in the text by omitting the kg-' from the MTX doses, and by referring to INDO 1.25 mg kg-' as INDO.
Comparison of mice with and without tumours Compared with mammary tissue from normal mice the tumours had less g-I of 7/17 FAs, more of 2/17 FAs and overall less combined amounts of n-3, total, unsaturated and saturated FAs.
Drug effects Treatment altered the amounts of tumour FAs, and the changes were often greater with MTX 4 mg than with MTX 2 mg (Table II). INDO alone also caused some Comparisons of tissues weights with vehicle-treated cancer-bearing mice are: ap <0.05; bp <0.02; cP <0.003. Normal mice had a median spleen weight of 72 mg which increased by 85% in the presence of tumour (to a median of 133 mg) and was almost normal (85 mg, 9.7% bigger) in mice given MTX 4 mg kg-' + INDO 1.25 mg kg-'. The lower median tumour and spleen weights with MTX 4 mg + INDO were not significantly different from those with MTX 4 mg alone. Normal mice had a median liver weight of 1.21 g. This was not significantly affected by the presence of tumour or the treatments administered. Vehicle-treated cancer group n = 9; vehicle-treated non-cancer group n = 12; other groups n = 6. increases, and the effect of MTX 2 mg ± INDO usually approximated to the sum of the changes obtained with the two drugs given separately. In contrast, INDO appeared to counteract the effect of MTX 4 mg. Most of the treatments reduced the ratio of tumour saturated/unsaturated FAs compared with the control tumours, because the amounts of unsaturated FAs tended to increase more than the saturated FAs. Combined amounts of the n-6 polyunsaturated FAs (18:2, 20:2, 20:3, 20:4, 22:2 and 22:4) increased with MTX 2 mg ± INDO or MTX 4 mg alone. The n-6/n-3 ratio also increased, because the combined amounts of the n-3 FAs (18:3, 18:4, 20:5 and 22:6) were not significantly changed.

Mammary fatty acids
These results are shown in Table III. Comparison of mice with and without tumours Amounts of FAs and the various ratios examined in the mammary tissue from tumour-bearing vehicle-treated and normal mice were similar.

Plasma fatty acids
These results are shown in Table IV. Comparison of mice with and without tumours Plasma from the cancer-bearing mice had smaller amounts of 4/17 FAs (14:0, 18:0, 18:4, 20:0) and more 18:1 and 18:3. Both groups contained similar amounts of total plasma FAs, but the tumour-bearers had less saturated FAs.
Drug effects In the untreated cancer group, amounts of plasma 14:0, 18:0, 18:4 and 20:0 were below normal, whereas 18:1 and 18:3 were raised. MTX 2 mg ± INDO in general appeared to inhibit the falls in the unsaturated FAs, and they increased the saturated/unsaturated and the 18:0/ 18:1 ratios. In contrast, MTX 4 mg ± INDO did not 'protect' against the cancer-induced falls, and actually reduced the amounts of several FAs (36% less total FAs; 29-41% less unsaturated FAs). All treatments increased the 16:0/16:1 ratio, but otherwise INDO usually caused no change.
Drug effects The total amounts of FAs extracted from the liver were similar in the treatment and control groups. No treatment counteracted the depression of FAs by the tumour, and any statistically significant changes of combined amounts or ratios were small.
or no change in 'normal' mammary tissue (the site of tumour origin ;Hewitt et al., 1976).
Fatty acid changes and the anticancer effect of cytotoxic drugs FA changes can affect anticancer therapy, and vice-versa. Cells enriched with polyunsaturates accumulated more adriamycin and MTX (Burns et al., 1979;Burns & North, 1986), and effective cytotoxic drugs caused an overall rise in the unsaturated FA content of cells (Schlager et al., 1980b). In our experiments MTX increased the tumour content of unsaturated FAs, and this effect might alter the cell membrane permeability and thickness (Schlager & Ohanian, 1980a,b).
Methotrexate/indomethacin interaction The MTX/INDO interaction is important because INDO potentiates both the MTX-induced prolongation of survival of mice with NC tumours, and the killing of NC cells and human breast cancer cells in culture (Bennett et al., 1987). The mechanism(s) are not fully understood, but we recently found that INDO potentiated the changes in FA composition induced by MTX in cultured NC cells (Soydan et al., 1991). However, potentiation rarely occurred in the present in vivo experiments.
Our previous results in vitro indicate that the effect does not involve MTX displacement from binding sites on serum proteins, or inhibition of prostaglandin formation, cAMP phosphodiesterase or of calcium transport (Gaffen et al., 1985(Gaffen et al., , 1989Bennett et al., 1987;Gaffen et al., 1991). However, inhibition of prostaglandin synthesis seems to explain the prolongation of survival by INDO in NC tumour-bearing mice , and we have not excluded the possibility that this mechanisms may contribute to the potentiation of MTX cytotoxicity in vivo. The spleen can synthesise large amounts of prostanoids such as PGE2, PGI2 and thromboxane A2 (Pace-Asciak & Rangaraj, 1977;Hidaka et al., 1983), and these prostanoids might affect the host response to the tumour (Bennett, 1982). In the NC tumour and spleen, amounts of 20:4 (the precursor of the 2-series prostaglandins) increased somewhat with MTX 2 mg. Perhaps the potentiation of MTX cytotoxicity by INDO in vivo (Bennett et al., 1987) involves a decrease in the formation of immunosuppressive PGE2, particularly since MTX itself causes immunosuppression (Jackson, 1984;Chabner et al., 1985), and cytotoxic drugs can increase prostaglandin release (Levine, 1977;Berstock et al., 1980). Prostaglandins are not the only lipids that can influence the immune system, and linoleate alone or in metabolic relationships with arachidonate and prostaglandins might be involved (Plescia et al., 1975). Mammary tumour cells synthesise primarily 18:3, 20:3 and 20:4 FAs from 18:2 (Chapkin et al., 1989), indicating the presence of desaturase and elongase enzymes. In our cancer-bearing mice, MTX 4 mg + INDO decreased the 18:2/20:4 ratio in the spleen, but increased it in the tumour. These results might reflect changed enzymic activities and/or prostaglandin production (Fulton, 1984;Hubbard et al., 1988).

Cachexia
The cachexia of malignancy is associated with weight loss and changes in body biochemistry which appear to be tumour-driven. Unlike starvation in a non-tumour-bearing host, the condition does not respond to 'corrective' nutrition. FA metabolism is involved, but the extent of this derangement is not known. The changes of tissue FAs that we obtained in response to the tumour and to therapy may be relevant to cancer cachexia.
In conclusion, FA changes occurred not only in the NC tumour compared to the normal mammary tissue from the same strain of mice in which it originally arose several years ago, but also in 'normal' tissues of cancer-bearing mice. The tumour changes relate in unexplained ways to carcinogenesis, and the 'normal' tissue FA alterations might relate to cachexia. It seems that some of these changes are reduced by treatment vith MTX ± INDO, particularly with the lower MTX dose of 2 mg kg-'. The amounts of FAs (ug g-, given to three significant figures) are shown as median values with interquartile ranges in parentheses. The treatment groups (n = 6, including the normal mice without tumour) are expressed as percentages of the vehicle-treated tumour-bearing controls (n = 9). P values <0.1, b<0.05, C<0.02, d< 0.002. 20:2 was not detected (ND) in any of the groups examined. Fatty acid ratios are shown at the bottom of the table.